Differences

This shows you the differences between two versions of the page.

--- doku:multimpi [2019/03/12 12:51] – jz
+++ doku:multimpi [2022/11/04 10:45] – [With srun:] goldenberg
@@ Line 3: / Line 3: @@
 Sample job script to run two MPI tasks within one job script concurrently:
-== With srun: ==
+==== VSC-4 =====
-<code>
-#!/bin/bash
-#SBATCH -J test
-#SBATCH -N 1
-#SBATCH --ntasks-per-core=1
-#SBATCH --ntasks-per-node=2
-export SLURM_STEP_GRES=none
+Sample Job when for running multiple mpi jobs on a VSC-4 node.
-module load intel/18 intel-mpi/2018
+Note: The "mem_per_task" should be set such that
-for i in 0 8
+<code>
-do
+mem_per_task * mytasks < mem_per_node - 2Gb
-  j=$(($i+1))
+</code>
-  srun -n 2 --cpu_bind=map_cpu:$i,$j ./hello_world_intelmpi2018 &
-done
-wait
-exit 0
+The approx 2Gb reduction in available memory is due to operating system stored in memory. For a standard node with 96 Gb of Memory this would be eg.:
+<code>
+Gb * 4 = 92 Gb < 94 Gb
 </code>
-== With mpirun (Intel MPI): ==
 <code>
 #!/bin/bash
-#SBATCH -J test
+#SBATCH -J many
 #SBATCH -N 1
-#SBATCH --ntasks-per-core=1
-#SBATCH --ntasks-per-node=2
 export SLURM_STEP_GRES=none
-module load intel/18 intel-mpi/2018
+mytasks=4
+cmd="stress -c 24"
+mem_per_task=10G
-for i in 0 8
+for i in `seq 1 $mytasks`
 do
-  j=$(($i+1))
+        srun --mem=$mem_per_task --cpus-per-task=2 --ntasks=1 $cmd &
-  mpirun -env I_MPI_PIN_PROCESSOR_LIST $i,$j -np 2 ./hello_world_intelmpi2018 &
 done
 wait
-exit 0
 </code>
-You can download the C code example here: {{ :doku:hello_world.c |hello_world.c}}
+==== VSC-3 =====